This study contributes to the better understanding of conditional cash transfer programs in several ways. It offers an empirical quantification of the effects on poverty of Oportunidades, an important program for Mexico and for other Latin American countries that have implemented similar programs; it provides an interesting benchmark for discussing the policy implications of how to implement this kind of programs and the political economy around it; and it provides an empirical example of the advantages and limitations of a specific methodology for ex ante program evaluations.
I start the discussion with the authors' quantification of the effects of Oportunidades on poverty. The paper assesses what would have happened to poverty under several scenarios: if the program had been cancelled, if its benefits were doubled under current targeting, if urban beneficiaries were doubled under current selection, if all poor were perfectly targeted by the program under current benefits, and if all poor were perfectly targeted by the program under doubled benefits. The authors estimate the economic costs of the program under each of these scenarios. The questions they ask are relevant to other Latin American and the Caribbean countries, many of which already have conditional transfer programs or might be considering them. To address these questions, the authors use the 2002 ENIGH survey to estimate an accounting exercise, which does not consider potential behavioral responses, and then estimate a behavioral model of those potential responses. Some of the key results of the accounting exercise are summarized in figure 4.
The figure illustrates the changes in the number of households below the poverty line and poverty gap, by policy or program design and by region. Panel A illustrates the status quo, in which the program is performing under the current mechanism of beneficiaries selection (CS) and current amount of benefits (CB): the x axis identifies how many poor households leave poverty, and the y axis shows how many dollars are required per month per household that leaves poverty, under different program designs. For example, the [End Page 109]
Click for larger view
View full resolution
[End Page 110]
program, as it is currently being implemented in the rural area (CS, CB in the figure), allows 387,000 households to leave poverty, at an average monthly cost of US$443 per poor household that leaves poverty. Perfectly targeting the current benefits given to all poor (AP, CB in the figure) would increase the number of households raised out of poverty to 579,000, at a monthly cost of US$231 per poor household. Doubling benefits under the current beneficiary selection mechanism (CS, DB in the figure) would result in 939,000 households leaving poverty, at a monthly cost of US$365 per household. The implications are similar for panel B, which illustrates the monthly cost in millions of U.S. dollars per one percent decrease in the poverty gap in the y axis versus the change in the poverty gap in the x axis. In both panels, a comparison of the results obtained under (AP, CB) with those obtained under (CS, DB) clearly illustrates that under any well behaved preference function, it would be better in urban areas to improve targeting rather than increase benefits.1 In rural areas, some preference functions might conceivably favor improving targeting over increasing benefits. However, given that reaching all poor households is basically unfeasible, increasing benefits provides more room to reduce both households under the poverty line and the poverty gap than does improving targeting.2
The accounting exercise, despite its simplicity, provides immediate directions in which the program should be reoriented. As the authors conclude, increasing the amount of transfers would be very cost effective in rural areas and worthless in urban communities, while improving the program's targeting would have a relatively much larger effect in the (much worse targeted) urban areas.
The second point mentioned above is that the paper provides an excellent benchmark for discussing cash transfer programs. It highlights deep differences in the implementation of a program in rural versus urban areas, which generated strikingly different results in terms of the program...